Mentoring is widely considered an crucial part of scientist training and access to good mentoring is through to be a limiting factor for the scientist pipeline from under-represented groups (URG). In particular, the Ibarra-Thomas Hypothesis from the business world suggests that URG groups need to receive psychosocial mentoring from mentors of their own group. Despite the extensive 'best practice' literature related to scientist training, there is little empirical understanding of the structure and dynamics of the social networks that mediate the developmental functions of mentoring. These limitations severely constrain our ability to evaluate mentoring and therefore our ability to make changes in the access to mentoring to enhance the URG scientist pipeline. The proposed research brings together biomedical researchers invested in mentoring of URG students and psychologists and sociologists who are experts in the fields of diversity, professional human relations and the development of human capital. We will test the hypothesis that access to specific forms of diverse mentoring are needed for effective instrumental and psychosocial mentoring. Aim 1 is to use study the structure and dynamics of personal mentoring networks used by 300 nationally distributed majority and URG biomedical PhD students. A novel microblogging approach to event-contingent reporting will be used to generate about 300,000 event reports recording 3 million values about mentoring interactions. These data will be combined with on- line questionnaires and video-interviews to build a dynamic picture of mentor-protg interactions and to test the Ibarra-Thomas Hypothesis and the more general diversity hypothesis to ask: what kinds of mentoring are needed and with whom? Aim 2 is to construct a data-based model of mentoring environments available at the US institutions training majority and URG students. Aim 3 is to conduct interviews with 150 nationally distributed research advisors to determine their attitudes and personal concepts of 'best practice' for mentoring students and URG students. These Aims will allow us to develop a comprehensive evidence-based model of how majority and URG biomedical PhD students are mentored in the United States.